SSP Technical Program
Thomas (Tom) J. Bzik, Versum Materials, Inc.
“Detecting the Detection Limit”
Detection limits have proven to be highly controversial in practice despite decades of developmental efforts by analytical scientists and regulators. This high resistance to obtaining effective and accepted detection limit solutions indicates our collective failure in how this problem is traditionally approached. The analytical chemist approaches it from analytical science, the statistician from statistical theory, the regulator from their a desire for regulatory needs, the instrument manufacturer from the standpoint of selling instruments, the contract laboratory from the standpoint of selling services, and the regulated from the standpoint of minimizing their reducing the cost of regulation. This exquisite mix of science, misapplication of science, regulatory needs, and business concerns screams part of the reason highlights why serious trouble abounds. Figuring out how to simultaneously address these constraints is not a trivial task.
Our journey of navigating the detection limit obstacle course successfully begins with a broad view of the problem focused on developing a conceptual understanding of key detection limit concepts and limitations. Throughout the presentation, key detection limit concepts, including the problems caused by our natural biases, will be explored. Issues in detection limit and rule-set application will be probed since it is at the application level that detection limit standards succeed or fail.
Tom currently works as a statistical consultant for Versum Materials, Inc. after an extended career at Air Products and Chemicals, Inc. He is Chair of ASTM – E11 Quality and Statistics which manages an extensive portfolio of international statistical standards. He chaired the Statistical Methods Task Force in SEMI International Standards for over 20 years. Tom has taught classes at several universities including his alma mater the University of Connecticut. He was the keynote speaker at ASTM’s Symposium for Detection Limits in 2018. Tom has extensively published papers and presentations with a focus on properly handling the unique data characteristics of trace contamination data. He considers difficult problems to solve “fun”.